29.01.2013 Views

Heiss W.D. (ed.) Quantum dots.. a doorway to - tiera.ru

Heiss W.D. (ed.) Quantum dots.. a doorway to - tiera.ru

Heiss W.D. (ed.) Quantum dots.. a doorway to - tiera.ru

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

168 C.W.J. Beenakker<br />

The τE-dependence of the gap obtain<strong>ed</strong> by Vavilov and Larkin is plott<strong>ed</strong><br />

in Fig. 21 (dash<strong>ed</strong> curve). It is close <strong>to</strong> the result of the effective RMT<br />

(solid curve). The two theories pr<strong>ed</strong>ict the same limit Egap → π�/2τE for<br />

τE/τdwell →∞. The asymp<strong>to</strong>tes given in [40] are<br />

Egap = γ5/2 �<br />

�<br />

�<br />

τE<br />

1 − 0.23 , τE≪τdwell , (92)<br />

τdwell 2τdwell<br />

Egap = π�<br />

�<br />

1 −<br />

2τE<br />

2τdwell<br />

�<br />

, τE≫τdwell . (93)<br />

τE<br />

Both are different from the results (88) and (89) of the effective RMT. 8<br />

8.4 Numerical Simulations<br />

Because the Ehrenfest time grows only logarithmically with the size of the<br />

system, it is exce<strong>ed</strong>ingly difficult <strong>to</strong> do numerical simulations deep in the<br />

Ehrenfest regime. Two simulations [27, 80] have been able <strong>to</strong> probe the initial<br />

decay of the excitation gap, when τE < ∼ τdwell. We show the results of both<br />

simulations in Fig. 24 (clos<strong>ed</strong> and open circles), <strong>to</strong>gether with the full decay as<br />

pr<strong>ed</strong>ict<strong>ed</strong> by the effective RMT of Sect. 8.2 (solid curve) and by the s<strong>to</strong>chastic<br />

model of Sect. 8.3 (dash<strong>ed</strong> curve).<br />

The clos<strong>ed</strong> circles were obtain<strong>ed</strong> by Jacquod et al. [27] using the stroboscopic<br />

model of Sect. 5 (the Andreev kick<strong>ed</strong> rota<strong>to</strong>r). The number of modes<br />

N in the contact <strong>to</strong> the superconduc<strong>to</strong>r was increas<strong>ed</strong> from 10 2 <strong>to</strong> 10 5 at<br />

fix<strong>ed</strong> dwell time τdwell = M/N = 5 and kicking strength K = 14 (corresponding<br />

<strong>to</strong> a Lyapunov exponent α ≈ ln(K/2) = 1.95). In this way all classical<br />

properties of the billiard remain the same while the effective Planck constant<br />

heff =1/M =1/N τdwell is r<strong>ed</strong>uc<strong>ed</strong> by three orders of magnitude. To plot the<br />

data as a function of τE/τdwell, (84) was us<strong>ed</strong> for the Ehrenfest time. The<br />

unspecifi<strong>ed</strong> terms of order unity in that equation were treat<strong>ed</strong> as a single fit<br />

parameter. (This amounts <strong>to</strong> a horizontal shift by −0.286 of the data points<br />

in Fig. 24.)<br />

The open circles were obtain<strong>ed</strong> by Kormányos et al. [80] for the chaotic<br />

Sinai billiard shown in the inset. The number of modes N was vari<strong>ed</strong> from<br />

18 <strong>to</strong> 30 by varying the width of the contact <strong>to</strong> the superconduc<strong>to</strong>r. The<br />

Lyapunov exponent α ≈ 1.7 was fix<strong>ed</strong>, but τdwell was not kept constant in<br />

this simulation. The Ehrenfest time was comput<strong>ed</strong> by means of the same<br />

formula (84), with M =2LckF /π and Lc the average length of a trajec<strong>to</strong>ry<br />

between two consecutive bounces at the curv<strong>ed</strong> boundary segment.<br />

The data points from both simulations have substantial error bars (up <strong>to</strong><br />

10%). Because of that and because of their limit<strong>ed</strong> range, we can not conclude<br />

that the simulations clearly favor one theory over the other.<br />

8 Since 2γ − 1=0.236, the small-τE asymp<strong>to</strong>te of Vavilov and Larkin differs by a<br />

fac<strong>to</strong>r of two from that of the effective RMT.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!